A self-taught AI engineer and amateur linguist says he has accomplished a feat that has eluded experts for more than a century: deciphering Linear A, the Bronze-Age writing system of the Minoan civilization. And he did it with the help of Claude Code, Anthropic's AI coding assistant.
Tom Di Mino, based in the Hudson Valley, claims that Linear A belongs to an extinct Semitic language that was a precursor to Biblical Hebrew, Arabic, and Aramaic. His work is currently under review by linguistics experts at Rutgers and Cambridge universities, according to a detailed account published on the AI Clambake blog. For more context on this story, see our ongoing more AI stories.
If verified, the decipherment would be an earthquake in the field of linguistics. When a closely related script, Linear B, was deciphered in 1952, it made the front page of The New York Times.
A 7-Year Obsession and a May Breakthrough
Di Mino has studied classical history, linguistics, and languages since he was 18 and has varying proficiency in eight languages, including Attic Greek, classical Latin, Sanskrit, Arabic, and Ugaritic. He has been reading about Linear A for seven years and has visited the Greek island of Crete twice.
He began actively working on deciphering Linear A in January 2026. According to his account, the major insight came to him on May 22, 2026, while analyzing a series of Linear A prayer inscriptions that followed a recognizable formula.
The formula appeared across five Minoan sanctuary sites. Every word in each line was already known from its overlap with Linear B — except the first word, which contained the same verb root in different regional forms. That word included a Linear A-only sign known as "*301." By connecting it to the Semitic consonant root N-W-Y, meaning "to dwell or inhabit," Di Mino says he unlocked the root "nawaya."
Once decoded, he found that the prayer resembled subsequent Hebrew prayers but was addressed to a goddess.
How Claude Code Changed the Equation
The role of AI in the project was practical rather than mystical. Di Mino used Claude Code to build a suite of Python scripts that query, cross-reference, and organize the digitized Linear A corpus, which is drawn from two major academic databases known as GORILA and SigLA.
These scripts enabled systematic hypothesis testing at a scale that would have been impractical to perform manually. By automating the tedious work of cross-referencing thousands of tablet transcriptions against proposed phonetic values, the tools let Di Mino rapidly test and refine his translations.
The approach echoes how Linear B was originally cracked. In that earlier case, Brooklyn College professor Alice Kober and the British architect Michael Ventris used grammatical and statistical analysis to find patterns in symbol placement and shifts. Di Mino's work extends that tradition into the era of computational linguistics, with AI doing the heavy lifting on data organization.
What the Decipherment Claims to Deliver
If accepted, Di Mino's findings would include several major contributions:
- 40 proposed readings for Linear A signs, including 13 signs whose phonetic values were previously unknown.
- Resolved sound values for five Linear B signs that remained unknown until now.
- A lexicon of 408 Linear A terms translated into English.
- A nine-page manuscript draft, titled Ya Diktu: Grammar of the Minoan Peak Sanctuary Libation Formula, which may form the basis for a submission to a peer-reviewed scientific journal.
Di Mino also claims that his insights into Linear A logograms help resolve problems with certain Linear B translations, which would independently validate his findings.
Not the First Semitic Hypothesis
Di Mino is not the first to argue that Linear A was Semitic. A 1957 article by Cyrus Gordon, published in the journal Antiquity, proposed links between Linear A dedication tablets and similar tablets in Akkadian and Phoenician that he had translated. Gordon's work, however, did not unlock translations the way Di Mino's solution appears to, and it never gained widespread acceptance.
What sets Di Mino's claim apart, he says, is that he is the first to identify links between Linear A inscriptions and Hebrew prayers specifically. That connection is what unlocked the verb in the prayer inscriptions and, potentially, broader patterns in the script's use of logograms.
A Steep Burden of Proof
Despite the excitement, the claim faces significant scrutiny. Linear A is notoriously difficult to decipher because most surviving inscriptions are inventory records cataloging the trade of commodities, which reveal little about the underlying language. There are also far fewer Linear A inscriptions than Linear B ones.
The 60 core syllables that Linear A shares with Linear B allowed experts to guess what the overlapping symbols sounded like, but not what the sounds meant. The 13 additional symbols unique to Linear A had no accepted sound values at all.
Di Mino's background as an amateur, while reminiscent of Ventris, also invites skepticism. The fact that his claims are under active review at Rutgers and Cambridge, rather than already published, underscores that the academic verdict is far from settled.
AI as a Tool for Ancient Mysteries
Regardless of whether Di Mino's decipherment holds up, his work points to a broader trend: AI tools are increasingly being turned toward problems in the humanities that have resisted conventional methods for decades or centuries.
In the Linear A case, Claude Code did not crack the script on its own. The linguistic insight — the link between the Minoan verb root and Semitic consonant patterns — came from Di Mino's years of study. What AI provided was the computational leverage to test that insight against a vast corpus of inscriptions at a speed no human could match.
As large language models and AI coding assistants become more capable, similar approaches may be applied to other undeciphered scripts, from the Indus Valley seals to the Rongorongo glyphs of Easter Island. The combination of human linguistic expertise and machine-scale pattern matching could open new frontiers in fields that once seemed frozen.
For now, the linguistics community will judge Di Mino's work on its merits. But the method behind it — an amateur scholar armed with deep language knowledge and an AI coding assistant — offers a glimpse of how artificial intelligence may help unlock the past, one ancient script at a time.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
